thread stage 3: create independant thread structure, unembed from proc.
[dragonfly.git] / sys / platform / pc32 / i386 / sys_machdep.c
CommitLineData
984263bc
MD
1/*-
2 * Copyright (c) 1990 The Regents of the University of California.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 * from: @(#)sys_machdep.c 5.5 (Berkeley) 1/19/91
34 * $FreeBSD: src/sys/i386/i386/sys_machdep.c,v 1.47.2.3 2002/10/07 17:20:00 jhb Exp $
1de703da 35 * $DragonFly: src/sys/platform/pc32/i386/sys_machdep.c,v 1.2 2003/06/17 04:28:35 dillon Exp $
984263bc
MD
36 *
37 */
38
39#include "opt_user_ldt.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/sysproto.h>
44#include <sys/malloc.h>
45#include <sys/proc.h>
46
47#include <vm/vm.h>
48#include <sys/lock.h>
49#include <vm/pmap.h>
50#include <vm/vm_map.h>
51#include <vm/vm_extern.h>
52
53#include <sys/user.h>
54
55#include <machine/cpu.h>
56#include <machine/ipl.h>
57#include <machine/pcb_ext.h> /* pcb.h included by sys/user.h */
58#include <machine/sysarch.h>
59#ifdef SMP
60#include <machine/smp.h>
61#endif
62
63#include <vm/vm_kern.h> /* for kernel_map */
64
65#define MAX_LD 8192
66#define LD_PER_PAGE 512
67#define NEW_MAX_LD(num) ((num + LD_PER_PAGE) & ~(LD_PER_PAGE-1))
68#define SIZE_FROM_LARGEST_LD(num) (NEW_MAX_LD(num) << 3)
69
70
71
72#ifdef USER_LDT
73static int i386_get_ldt __P((struct proc *, char *));
74static int i386_set_ldt __P((struct proc *, char *));
75#endif
76static int i386_get_ioperm __P((struct proc *, char *));
77static int i386_set_ioperm __P((struct proc *, char *));
78int i386_extend_pcb __P((struct proc *));
79
80#ifndef _SYS_SYSPROTO_H_
81struct sysarch_args {
82 int op;
83 char *parms;
84};
85#endif
86
87int
88sysarch(p, uap)
89 struct proc *p;
90 register struct sysarch_args *uap;
91{
92 int error = 0;
93
94 switch(uap->op) {
95#ifdef USER_LDT
96 case I386_GET_LDT:
97 error = i386_get_ldt(p, uap->parms);
98 break;
99
100 case I386_SET_LDT:
101 error = i386_set_ldt(p, uap->parms);
102 break;
103#endif
104 case I386_GET_IOPERM:
105 error = i386_get_ioperm(p, uap->parms);
106 break;
107 case I386_SET_IOPERM:
108 error = i386_set_ioperm(p, uap->parms);
109 break;
110 case I386_VM86:
111 error = vm86_sysarch(p, uap->parms);
112 break;
113 default:
114 error = EOPNOTSUPP;
115 break;
116 }
117 return (error);
118}
119
120int
121i386_extend_pcb(struct proc *p)
122{
123 int i, offset;
124 u_long *addr;
125 struct pcb_ext *ext;
126 struct soft_segment_descriptor ssd = {
127 0, /* segment base address (overwritten) */
128 ctob(IOPAGES + 1) - 1, /* length */
129 SDT_SYS386TSS, /* segment type */
130 0, /* priority level */
131 1, /* descriptor present */
132 0, 0,
133 0, /* default 32 size */
134 0 /* granularity */
135 };
136
137 ext = (struct pcb_ext *)kmem_alloc(kernel_map, ctob(IOPAGES+1));
138 if (ext == 0)
139 return (ENOMEM);
140 p->p_addr->u_pcb.pcb_ext = ext;
141 bzero(ext, sizeof(struct pcb_ext));
142 ext->ext_tss.tss_esp0 = (unsigned)p->p_addr + ctob(UPAGES) - 16;
143 ext->ext_tss.tss_ss0 = GSEL(GDATA_SEL, SEL_KPL);
144 /*
145 * The last byte of the i/o map must be followed by an 0xff byte.
146 * We arbitrarily allocate 16 bytes here, to keep the starting
147 * address on a doubleword boundary.
148 */
149 offset = PAGE_SIZE - 16;
150 ext->ext_tss.tss_ioopt =
151 (offset - ((unsigned)&ext->ext_tss - (unsigned)ext)) << 16;
152 ext->ext_iomap = (caddr_t)ext + offset;
153 ext->ext_vm86.vm86_intmap = (caddr_t)ext + offset - 32;
154
155 addr = (u_long *)ext->ext_vm86.vm86_intmap;
156 for (i = 0; i < (ctob(IOPAGES) + 32 + 16) / sizeof(u_long); i++)
157 *addr++ = ~0;
158
159 ssd.ssd_base = (unsigned)&ext->ext_tss;
160 ssd.ssd_limit -= ((unsigned)&ext->ext_tss - (unsigned)ext);
161 ssdtosd(&ssd, &ext->ext_tssd);
162
163 /* switch to the new TSS after syscall completes */
164 need_resched();
165
166 return 0;
167}
168
169static int
170i386_set_ioperm(p, args)
171 struct proc *p;
172 char *args;
173{
174 int i, error;
175 struct i386_ioperm_args ua;
176 char *iomap;
177
178 if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
179 return (error);
180
181 if ((error = suser(p)) != 0)
182 return (error);
183 if (securelevel > 0)
184 return (EPERM);
185 /*
186 * XXX
187 * While this is restricted to root, we should probably figure out
188 * whether any other driver is using this i/o address, as so not to
189 * cause confusion. This probably requires a global 'usage registry'.
190 */
191
192 if (p->p_addr->u_pcb.pcb_ext == 0)
193 if ((error = i386_extend_pcb(p)) != 0)
194 return (error);
195 iomap = (char *)p->p_addr->u_pcb.pcb_ext->ext_iomap;
196
197 if (ua.start + ua.length > IOPAGES * PAGE_SIZE * NBBY)
198 return (EINVAL);
199
200 for (i = ua.start; i < ua.start + ua.length; i++) {
201 if (ua.enable)
202 iomap[i >> 3] &= ~(1 << (i & 7));
203 else
204 iomap[i >> 3] |= (1 << (i & 7));
205 }
206 return (error);
207}
208
209static int
210i386_get_ioperm(p, args)
211 struct proc *p;
212 char *args;
213{
214 int i, state, error;
215 struct i386_ioperm_args ua;
216 char *iomap;
217
218 if ((error = copyin(args, &ua, sizeof(struct i386_ioperm_args))) != 0)
219 return (error);
220 if (ua.start >= IOPAGES * PAGE_SIZE * NBBY)
221 return (EINVAL);
222
223 if (p->p_addr->u_pcb.pcb_ext == 0) {
224 ua.length = 0;
225 goto done;
226 }
227
228 iomap = (char *)p->p_addr->u_pcb.pcb_ext->ext_iomap;
229
230 i = ua.start;
231 state = (iomap[i >> 3] >> (i & 7)) & 1;
232 ua.enable = !state;
233 ua.length = 1;
234
235 for (i = ua.start + 1; i < IOPAGES * PAGE_SIZE * NBBY; i++) {
236 if (state != ((iomap[i >> 3] >> (i & 7)) & 1))
237 break;
238 ua.length++;
239 }
240
241done:
242 error = copyout(&ua, args, sizeof(struct i386_ioperm_args));
243 return (error);
244}
245
246#ifdef USER_LDT
247/*
248 * Update the GDT entry pointing to the LDT to point to the LDT of the
249 * current process. Do not staticize.
250 */
251void
252set_user_ldt(struct pcb *pcb)
253{
254 struct pcb_ldt *pcb_ldt;
255
256 if (pcb != curpcb)
257 return;
258
259 pcb_ldt = pcb->pcb_ldt;
260#ifdef SMP
261 gdt[cpuid * NGDT + GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
262#else
263 gdt[GUSERLDT_SEL].sd = pcb_ldt->ldt_sd;
264#endif
265 lldt(GSEL(GUSERLDT_SEL, SEL_KPL));
266 currentldt = GSEL(GUSERLDT_SEL, SEL_KPL);
267}
268
269struct pcb_ldt *
270user_ldt_alloc(struct pcb *pcb, int len)
271{
272 struct pcb_ldt *pcb_ldt, *new_ldt;
273
274 MALLOC(new_ldt, struct pcb_ldt *, sizeof(struct pcb_ldt),
275 M_SUBPROC, M_WAITOK);
276 if (new_ldt == NULL)
277 return NULL;
278
279 new_ldt->ldt_len = len = NEW_MAX_LD(len);
280 new_ldt->ldt_base = (caddr_t)kmem_alloc(kernel_map,
281 len * sizeof(union descriptor));
282 if (new_ldt->ldt_base == NULL) {
283 FREE(new_ldt, M_SUBPROC);
284 return NULL;
285 }
286 new_ldt->ldt_refcnt = 1;
287 new_ldt->ldt_active = 0;
288
289 gdt_segs[GUSERLDT_SEL].ssd_base = (unsigned)new_ldt->ldt_base;
290 gdt_segs[GUSERLDT_SEL].ssd_limit = len * sizeof(union descriptor) - 1;
291 ssdtosd(&gdt_segs[GUSERLDT_SEL], &new_ldt->ldt_sd);
292
293 if ((pcb_ldt = pcb->pcb_ldt)) {
294 if (len > pcb_ldt->ldt_len)
295 len = pcb_ldt->ldt_len;
296 bcopy(pcb_ldt->ldt_base, new_ldt->ldt_base,
297 len * sizeof(union descriptor));
298 } else {
299 bcopy(ldt, new_ldt->ldt_base, sizeof(ldt));
300 }
301 return new_ldt;
302}
303
304void
305user_ldt_free(struct pcb *pcb)
306{
307 struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
308
309 if (pcb_ldt == NULL)
310 return;
311
312 if (pcb == curpcb) {
313 lldt(_default_ldt);
314 currentldt = _default_ldt;
315 }
316
317 if (--pcb_ldt->ldt_refcnt == 0) {
318 kmem_free(kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
319 pcb_ldt->ldt_len * sizeof(union descriptor));
320 FREE(pcb_ldt, M_SUBPROC);
321 }
322 pcb->pcb_ldt = NULL;
323}
324
325static int
326i386_get_ldt(p, args)
327 struct proc *p;
328 char *args;
329{
330 int error = 0;
331 struct pcb *pcb = &p->p_addr->u_pcb;
332 struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
333 int nldt, num;
334 union descriptor *lp;
335 int s;
336 struct i386_ldt_args ua, *uap = &ua;
337
338 if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
339 return(error);
340
341#ifdef DEBUG
342 printf("i386_get_ldt: start=%d num=%d descs=%p\n",
343 uap->start, uap->num, (void *)uap->descs);
344#endif
345
346 /* verify range of LDTs exist */
347 if ((uap->start < 0) || (uap->num <= 0))
348 return(EINVAL);
349
350 s = splhigh();
351
352 if (pcb_ldt) {
353 nldt = pcb_ldt->ldt_len;
354 num = min(uap->num, nldt);
355 lp = &((union descriptor *)(pcb_ldt->ldt_base))[uap->start];
356 } else {
357 nldt = sizeof(ldt)/sizeof(ldt[0]);
358 num = min(uap->num, nldt);
359 lp = &ldt[uap->start];
360 }
361 if (uap->start + num > nldt) {
362 splx(s);
363 return(EINVAL);
364 }
365
366 error = copyout(lp, uap->descs, num * sizeof(union descriptor));
367 if (!error)
368 p->p_retval[0] = num;
369
370 splx(s);
371 return(error);
372}
373
374static int
375i386_set_ldt(p, args)
376 struct proc *p;
377 char *args;
378{
379 int error = 0, i, n;
380 int largest_ld;
381 struct pcb *pcb = &p->p_addr->u_pcb;
382 struct pcb_ldt *pcb_ldt = pcb->pcb_ldt;
383 union descriptor *descs;
384 int descs_size, s;
385 struct i386_ldt_args ua, *uap = &ua;
386
387 if ((error = copyin(args, uap, sizeof(struct i386_ldt_args))) < 0)
388 return(error);
389
390#ifdef DEBUG
391 printf("i386_set_ldt: start=%d num=%d descs=%p\n",
392 uap->start, uap->num, (void *)uap->descs);
393#endif
394
395 /* verify range of descriptors to modify */
396 if ((uap->start < 0) || (uap->start >= MAX_LD) || (uap->num < 0) ||
397 (uap->num > MAX_LD))
398 {
399 return(EINVAL);
400 }
401 largest_ld = uap->start + uap->num - 1;
402 if (largest_ld >= MAX_LD)
403 return(EINVAL);
404
405 /* allocate user ldt */
406 if (!pcb_ldt || largest_ld >= pcb_ldt->ldt_len) {
407 struct pcb_ldt *new_ldt = user_ldt_alloc(pcb, largest_ld);
408 if (new_ldt == NULL)
409 return ENOMEM;
410 if (pcb_ldt) {
411 pcb_ldt->ldt_sd = new_ldt->ldt_sd;
412 kmem_free(kernel_map, (vm_offset_t)pcb_ldt->ldt_base,
413 pcb_ldt->ldt_len * sizeof(union descriptor));
414 pcb_ldt->ldt_base = new_ldt->ldt_base;
415 pcb_ldt->ldt_len = new_ldt->ldt_len;
416 FREE(new_ldt, M_SUBPROC);
417 } else
418 pcb->pcb_ldt = pcb_ldt = new_ldt;
419#ifdef SMP
420 /* signal other cpus to reload ldt */
421 smp_rendezvous(NULL, (void (*)(void *))set_user_ldt, NULL, pcb);
422#else
423 set_user_ldt(pcb);
424#endif
425 }
426
427 descs_size = uap->num * sizeof(union descriptor);
428 descs = (union descriptor *)kmem_alloc(kernel_map, descs_size);
429 if (descs == NULL)
430 return (ENOMEM);
431 error = copyin(&uap->descs[0], descs, descs_size);
432 if (error) {
433 kmem_free(kernel_map, (vm_offset_t)descs, descs_size);
434 return (error);
435 }
436 /* Check descriptors for access violations */
437 for (i = 0, n = uap->start; i < uap->num; i++, n++) {
438 union descriptor *dp;
439 dp = &descs[i];
440
441 switch (dp->sd.sd_type) {
442 case SDT_SYSNULL: /* system null */
443 dp->sd.sd_p = 0;
444 break;
445 case SDT_SYS286TSS: /* system 286 TSS available */
446 case SDT_SYSLDT: /* system local descriptor table */
447 case SDT_SYS286BSY: /* system 286 TSS busy */
448 case SDT_SYSTASKGT: /* system task gate */
449 case SDT_SYS286IGT: /* system 286 interrupt gate */
450 case SDT_SYS286TGT: /* system 286 trap gate */
451 case SDT_SYSNULL2: /* undefined by Intel */
452 case SDT_SYS386TSS: /* system 386 TSS available */
453 case SDT_SYSNULL3: /* undefined by Intel */
454 case SDT_SYS386BSY: /* system 386 TSS busy */
455 case SDT_SYSNULL4: /* undefined by Intel */
456 case SDT_SYS386IGT: /* system 386 interrupt gate */
457 case SDT_SYS386TGT: /* system 386 trap gate */
458 case SDT_SYS286CGT: /* system 286 call gate */
459 case SDT_SYS386CGT: /* system 386 call gate */
460 /* I can't think of any reason to allow a user proc
461 * to create a segment of these types. They are
462 * for OS use only.
463 */
464 kmem_free(kernel_map, (vm_offset_t)descs, descs_size);
465 return EACCES;
466
467 /* memory segment types */
468 case SDT_MEMEC: /* memory execute only conforming */
469 case SDT_MEMEAC: /* memory execute only accessed conforming */
470 case SDT_MEMERC: /* memory execute read conforming */
471 case SDT_MEMERAC: /* memory execute read accessed conforming */
472 /* Must be "present" if executable and conforming. */
473 if (dp->sd.sd_p == 0) {
474 kmem_free(kernel_map, (vm_offset_t)descs,
475 descs_size);
476 return (EACCES);
477 }
478 break;
479 case SDT_MEMRO: /* memory read only */
480 case SDT_MEMROA: /* memory read only accessed */
481 case SDT_MEMRW: /* memory read write */
482 case SDT_MEMRWA: /* memory read write accessed */
483 case SDT_MEMROD: /* memory read only expand dwn limit */
484 case SDT_MEMRODA: /* memory read only expand dwn lim accessed */
485 case SDT_MEMRWD: /* memory read write expand dwn limit */
486 case SDT_MEMRWDA: /* memory read write expand dwn lim acessed */
487 case SDT_MEME: /* memory execute only */
488 case SDT_MEMEA: /* memory execute only accessed */
489 case SDT_MEMER: /* memory execute read */
490 case SDT_MEMERA: /* memory execute read accessed */
491 break;
492 default:
493 kmem_free(kernel_map, (vm_offset_t)descs, descs_size);
494 return(EINVAL);
495 /*NOTREACHED*/
496 }
497
498 /* Only user (ring-3) descriptors may be present. */
499 if ((dp->sd.sd_p != 0) && (dp->sd.sd_dpl != SEL_UPL)) {
500 kmem_free(kernel_map, (vm_offset_t)descs, descs_size);
501 return (EACCES);
502 }
503 }
504
505 s = splhigh();
506
507 /* Fill in range */
508 bcopy(descs,
509 &((union descriptor *)(pcb_ldt->ldt_base))[uap->start],
510 uap->num * sizeof(union descriptor));
511 p->p_retval[0] = uap->start;
512
513 splx(s);
514 kmem_free(kernel_map, (vm_offset_t)descs, descs_size);
515 return (0);
516}
517#endif /* USER_LDT */